t-DGR: A Trajectory-Based Deep Generative Replay Method for Continual Learning in Decision Making

William Yue, Bo Liu, Peter Stone
Proceedings of The 3rd Conference on Lifelong Learning Agents, PMLR 274:481-497, 2025.

Abstract

Deep generative replay has emerged as a promising approach for continual learning in decision-making tasks. This approach addresses the problem of catastrophic forgetting by leveraging the generation of trajectories from previously encountered tasks to augment the current dataset. However, existing deep generative replay methods for continual learning rely on autoregressive models, which suffer from compounding errors in the generated trajectories. In this paper, we propose a simple, scalable, and non-autoregressive method for continual learning in decision-making tasks using a generative model that generates task samples conditioned on the trajectory timestep. We evaluate our method on Continual World benchmarks and find that our approach achieves state-of-the-art performance on the average success rate metric among continual learning methods.

Cite this Paper


BibTeX
@InProceedings{pmlr-v274-yue25a, title = {t-DGR: A Trajectory-Based Deep Generative Replay Method for Continual Learning in Decision Making}, author = {Yue, William and Liu, Bo and Stone, Peter}, booktitle = {Proceedings of The 3rd Conference on Lifelong Learning Agents}, pages = {481--497}, year = {2025}, editor = {Lomonaco, Vincenzo and Melacci, Stefano and Tuytelaars, Tinne and Chandar, Sarath and Pascanu, Razvan}, volume = {274}, series = {Proceedings of Machine Learning Research}, month = {29 Jul--01 Aug}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v274/main/assets/yue25a/yue25a.pdf}, url = {https://proceedings.mlr.press/v274/yue25a.html}, abstract = {Deep generative replay has emerged as a promising approach for continual learning in decision-making tasks. This approach addresses the problem of catastrophic forgetting by leveraging the generation of trajectories from previously encountered tasks to augment the current dataset. However, existing deep generative replay methods for continual learning rely on autoregressive models, which suffer from compounding errors in the generated trajectories. In this paper, we propose a simple, scalable, and non-autoregressive method for continual learning in decision-making tasks using a generative model that generates task samples conditioned on the trajectory timestep. We evaluate our method on Continual World benchmarks and find that our approach achieves state-of-the-art performance on the average success rate metric among continual learning methods.} }
Endnote
%0 Conference Paper %T t-DGR: A Trajectory-Based Deep Generative Replay Method for Continual Learning in Decision Making %A William Yue %A Bo Liu %A Peter Stone %B Proceedings of The 3rd Conference on Lifelong Learning Agents %C Proceedings of Machine Learning Research %D 2025 %E Vincenzo Lomonaco %E Stefano Melacci %E Tinne Tuytelaars %E Sarath Chandar %E Razvan Pascanu %F pmlr-v274-yue25a %I PMLR %P 481--497 %U https://proceedings.mlr.press/v274/yue25a.html %V 274 %X Deep generative replay has emerged as a promising approach for continual learning in decision-making tasks. This approach addresses the problem of catastrophic forgetting by leveraging the generation of trajectories from previously encountered tasks to augment the current dataset. However, existing deep generative replay methods for continual learning rely on autoregressive models, which suffer from compounding errors in the generated trajectories. In this paper, we propose a simple, scalable, and non-autoregressive method for continual learning in decision-making tasks using a generative model that generates task samples conditioned on the trajectory timestep. We evaluate our method on Continual World benchmarks and find that our approach achieves state-of-the-art performance on the average success rate metric among continual learning methods.
APA
Yue, W., Liu, B. & Stone, P.. (2025). t-DGR: A Trajectory-Based Deep Generative Replay Method for Continual Learning in Decision Making. Proceedings of The 3rd Conference on Lifelong Learning Agents, in Proceedings of Machine Learning Research 274:481-497 Available from https://proceedings.mlr.press/v274/yue25a.html.

Related Material